Final discussion

5.1.1 ELF3 and GI collaborate in thermal signal input pathway

I identified that the hsp90.2-3 mutant and GDA-treated seedlings showed a longer period phenotype (Figure 3.1 and 3.2). I further identified ELF3 is one client of Hsp90. Once Hsp90.2 in the cell was inhibited by GDA, the amount of ELF3 was reduced (Figure 3.11, 3.12, and 3.13). Transcription levels of PRR7 and PRR9 were not significantly elevated by GDA treatment (Figure 3.7). However, it was demonstrated that ELF3 is the transcription repressor of PRR9. In the elf3 mutant, the gene accumulation level of PRR9 was elevated. Meanwhile, the period of weak alleles of elf3 was shortened (Kolmos et al., 2011). In addition, the gi mutant was still influenced by GDA. However, it has been proposed that Hsp90 influences the clock through ZTL, a downstream protein regulated by GI. Hsp90 and GI are tightly connected in the ZTL-stabilization, as the low levels of ZTL in gi mutant is not significantly reduced by GDA, which suggests that GDA should not result in a longer period in the gi mutant (Kim et al., 2007).

To explain these conflicts, I propose that both ELF3 and GI are involved in the temperature input pathway. These two components may be involved in two partially independent pathways. ELF3 and GI influence transcription of CCA1 and LHY in opposite ways. ELF3 represses the transcription of PRR9 whereas PRR9 represses the transcription of CCA1 and LHY (Herrero et al., 2012; Nakamichi et al., 2005).

Therefore, in theory, ELF3 positively regulates the transcription of CCA1 and LHY.

However, GI was demonstrated to positively regulate the transcription of CCA1 and LHY. The gi mutations result in a reduction in the expression of CCA1 and LHY (Fowler et al., 1999). Meanwhile, the gi mutants cause a short-period phenotype (Park et al., 1999). Interestingly, the nighttime repressor ELF3-ELF4-LUX negatively regulates the transcription of GI (Mizuno et al., 2014). Thus, the accumulation level of CCA1 and LHY may reach the balance between the regulations of ELF3 and GI. In conclusion, Hsp90 influences the circadian clock through both ELF3 and GI, and ELF3 and GI are involved in partially independent pathways.

5.1.2 PRR9 and CCA1 may serve as stress indicators

Under normal conditions, Hsps are mainly located in the cytoplasm. However, under stress conditions, Hsps rapidly transfer to the nucleus (Horwitz, 1992; Lindquist and Craig, 1988). When HS90.2 was expressed alone in tobacco cells, it was found in the cytoplasm. However, Hsp90.2 was co-expressed with 35s::PRR9, some Hsp90.2 transferred to the nucleus (Figure 3.12A), which suggested that accumulation of PRR9 may be recognized as a signal of stress.

The gene accumulation of PRR9 was elevated in warm conditions (Mizuno et al., 2014). Considering this point, it is reasonable to match the expression of PRR9 to the daily cycle. Normally, temperature increases from early morning to the middle of the day. Meanwhile, expression of PRR9 gradually increases in the morning and reaches its peak level in the middle of the day. The oscillation of PRR9 accumulation perfectly matches the daily rhythmic changes of temperature. The average accumulation level of PRR9 may match the seasonal changes, as in summer the average level of PRR9 would be higher than that in winter. Therefore, my hypothesis is that PRR9 could serve as a “thermometer” which tells the plant the temperature. Overexpression of PRR9 may mislead the plant in sensing the actual temperature, simulating a heat stress and triggering stress responses, which is like a “fever”.

Figure 5.1 CCA1-GFP is co-expressed with Hsp90.2-RFP in N. benthamiana.

When CCA1 was highly expressed in the nucleus, Hsp90.2 transferred into nucleus but is not completely co-localized with CCA1. The green channel is GFP tagged CCA1. The red channel is RFP tagged Hsp90.2

CCA1 may be also recognized as another stress indicator. Hsp90.2 transferred to the nucleus when CCA1 was highly expressed (Figure 5.1). It appears that CCA1 was involved in the biotic stress regulation pathway. When I injected transformed agrobacteria with several clock constructs into tobacco leaves, only the Agrobacterium containing 35s::CCA1 resulted in a severe infection, which might indicate that CCA1 is involved in the plant immune system. However, it was previously demonstrated that CCA1 controls the expression of defense genes and timing of the immune response. The cca1 mutant showed compromised resistance whereas the CCA1-overexpression line showed enhanced resistance (Wang et al., 2011). This is completely opposite to my observation. One explanation is that the exogenous CCA1 strongly competes with native CCA1 in tobacco, which heavily

to pathogens. If considering the low expression level of CCA1 at higher temperature, there should be a “clock-temperature-immune triangle” in Arabidopsis.

5.1.3 Temperature alters the functions of evening complex ELF3-ELF4-LUX CCA1, LHY, PRR7, PRR9, and GI respond to a temperature upshift only during the dark period. The transcription of these genes were regulated by the evening complex (EC) night repressor ELF3-ELF4-LUX. A warmer temperature inhibits EC function, whereas a cooler temperature stimulates its function (Mizuno et al., 2014).

It is still unclear whether DNA-binding of EC to the target promoters is inhibited by a warm temperature or a warm temperature inhibits the repressor ability of EC.

Considering the role of Hsp90, it can either regulate the protein stability of ELF3 or correct formation of EC. When Hsp90 is not sufficient to stabilize the proteins, the ability of EC may be inhibited under a warm condition.

5.1.4 Hypothesis on CCA1/LHY-PRR7/PRR9 self-balancing loop

CCA1/LHY activates transcription of PRR7/PRR9 whereas PRR7 inhibits transcription of CCA1/LHY (Nakamichi et al., 2005). When the accumulation of CCA1 and LHY were reduced from ZT20 to ZT2 by GDA, the accumulation of PRR7 and PRR9 also increased at a slower pace (Figure 3.6 and 3.7). After ZT4, the lower accumulation of PRR7 and PRR9 in GDA-treated seedlings did not inhibit CCA1 and LHY as much as in the non-treated seedlings. Therefore, the accumulation of CCA1 and LHY became relatively higher after ZT4. Meanwhile, expression of PRR7 and PRR9 after ZT10 was also elevated, which might be due to the higher amount of CCA1 and LHY. It was noted that the expression peaks of CCA1/LHY and PRR7/PRR9 were all shifted (Figure 3.6 and 3.7).

As mentioned in the discussion part of Chapter three, the effective expression window of CCA1/LHY was widened by GDA, which may explain the long period phenotype. In fact, less active Hsp90 may make the clock more sensitive to temperature. GDA may degrade clock-related components at 22°C as fast as that at 30°C, and thus switched the seedlings into a “summer state”. Based on this hypothesis, the wider effective expression window of CCA1/LHY matches the

long-peak around 2 p.m., which is around 8 hours after dawn. This point matches the shifted expression pattern of PRR7/PRR9.

In summary, my hypothesis is that the CCA1/LHY-PRR7/PRR9 morning loop model may tell us how the plant anticipates the daily rhythmic changes in different seasons.

5.1.5 three-layer clock model

So far, developed mathematical models consist of the core loop, the morning loop and the evening loop. However, one question remains to be answered: which component contributes more to the clock? Here I propose a modified model, in which the clock components are classified into three hierarchies.

The first layer (inner layer) contains three core components: CCA1, LHY, and TOC1 (Figure 5.2 ). These three components define the “clock”. These three genes are notably crucial, as in the cca1 lhy toc1 triple mutant, the clock stays arrhythmic (Green and Tobin, 1999; Mizoguchi et al., 2002). Moreover, overexpression of either CCA1 or LHY causes arrhythmicity of the clock (Schaffer et al., 1998; Wang and Tobin, 1998). The second layer (middle layer) contains the “adjusters” (Figure 5.2).

The signal receivers are the outputs of clock. They form interlocked loops with core components, which reciprocally regulate each other. Once they receive the signals from the third layer (outer layer), their own expression or function are altered. They form interlocked loops with core clock components, thus the core clock is finely adjusted. PRR7 and PRR9 are two of the adjusters. The prr7, prr9, and the double mutant prr7 prr9, make the clock slower (Alabadi et al., 2001; Farre et al., 2005;

Yamamoto et al., 2003). In addition, the prr7-3 prr9-1 double mutant is not able to be reset by temperature pulses (Salome and McClung, 2005; Salome et al., 2010). The third layer (outer layer) contains the “signal mediators” (Figure 5.2). The signal mediators interact with the signal receivers to complete the signal transduction. The signal mediators do not have to be clock outputs. ELF3 may be one of the mediators.

The repressing efficiency of ELF3 on PRR7 and PRR9 changes at different temperatures (Mizuno et al., 2014). The temperature induction of PRR7 and PRR9 mRNA accumulations were eliminated in the elf3-1 mutant. (Thines and Harmon,

Figure 5.2 The three-layer clock model.

The three-layer clock model divides the clock into three hierarchies. The core components CCA1, LHY, and TOC1 are included in the first layer (white cycle). Defined as “adjuster”, PRR7 and PRR9 are included in the second layer (light green cycle). The third layer (dark green cycle) contains the “signal mediators”, such as ELF3

In conclusion, since many organisms have evolved a circadian clock, the clock itself should be fairly conserved. Therefore, I propose that the clock only contains two genes, a MyB and a PRR. Considering the algae, its clock may not contain a temperature-regulation pathway, due to constant temperature in water. From an evolutionary perspective, living organisms may develop temperature-regulating clock systems when they left the water and started to live on the land. This could have driven an increase in clock complexity.